On Dec 11, 2003, at 9:55 PM, Andy Adler wrote:
>>> In octave imread returns a greyscale image even when you read color
>>> images if you only have one variable to catch the return value. In
>>> matlab, you actually get a 3D array of size y,x,3 if you only specify
>>> one variable.
>>>
>>> Can this be changed?
>>>
>> It can now that Octave supports 3D arrays. Anyone interested
>> in implementing it?
>
> I could take a stab at it in a few weeks. However
>
> 1. Is the current imread implementation OK?
> ie. are we satisfied with piping stuff through the
> imagemagick 'convert'?
It saves us work if convert is there. Is it much slower than
using readjpg or readpng directly?
>
> 2. What is octave-forge supposed to support?
>
> N-d arrays are experimental, so if we modify the code,
> the octave-forge won't work for all those people that
> we're recommending should use 2.1.50.
At configure time we can create the function has_nD.m which
returns either true or false, then test at run-time what we want
to return.
- Paul
|